IBM CEO Arvind Krishna addresses a panel discussion at the World Economic Forum in Davos, Switzerland, January 17, 2023.
Stefan Wermuth | Bloomberg | Getty Images
It’s been a long time since IBM actively touted Watson. Originally created to protect people in the “Jeopardy!” Game show marked Watson IBM’s early leap into artificial intelligence, but it never materialized into a profitable offering.
About 15 months ago, IBM sold its Watson Health division to private equity firm Francisco Partners for an undisclosed amount.
Now Watson has given way to WatsonX, and IBM is trying to ride the recent AI boom. IBM bills it as an enterprise development studio for “training, tuning, and deploying” machine learning models. The platform includes an AI-generated code capability, an AI governance toolkit, and a library of thousands of large-scale AI models trained on speech, geospatial data, IT events, and code, according to a press release.
The new offering is part of a larger shift in strategy as IBM seeks to take the lead in easy-to-use platforms for companies looking to introduce AI into their business models, in part because there is a massive shortage of human talent in the AI market.
IBM partners with HuggingFace, the buoyant AI startup and open-source platform that hit $2 billion last year.
In a question-and-answer session with reporters Monday, IBM CEO Arvind Krishna said companies can come in with a model they want to build and then let WatsonX do the work.
“We allow a company to use their own code to customize the model to fit the way they want their playbooks and code to run,” Krishna said. “Then they can deploy it for themselves without the risk of their code leaking.”
Customers and collaborators so far include SAP, NASA, Wix and PyTorch, and the platform will be available in the third quarter, IBM told CNBC. The company declined to share how expensive it was to build WatsonX or how long it took. Last month, IBM’s quarterly earnings beat analysts’ expectations, although revenue fell short of estimates.
Krishna said he expects these new AI tools to be most easily integrated into areas such as customer care, procurement, cybersecurity, and elements of supply chain and IT operations. In particular, they will replace “repetitive back-office processes,” he said.
“We see that easily being between 30 and 50% of that volume of work and being able to do it with really as much or better competence than even humans can,” Krishna said. “We see this lot being accepted immediately starting this year and unfolding its full impact over the next three to five years.”
REGARD: AI will destroy jobs but ultimately create “massive amounts of wealth,” says the investment firm